Method for selecting an acquisition geometry for a seismic survey based on ability to resolve an a priori velocity model. Two or more candidate acquisition geometries are selected (301, 302), differing in areal coverage and cost to perform. Then compute a synthetic seismic dataset for each geometry using a detailed geometrical reference model of the subsurface (301). Invert the synthetic seismic datasets preferably using simultaneous source FWI, and preferably with Volume of Investigation constraints, to determine model updates (303, 304). Quantitatively assess the value of the additional traces in a fuller dataset relative to a subset (306), using one or more statistics based on the accuracy of the updated models, such as improved match to the reference model, better fit of seismic data, or rate of change in improvement with iterations. Inversions may be cascaded for further efficiency (314).
Legal claims defining the scope of protection, as filed with the USPTO.
1. A method for evaluating seismic survey acquisition geometries, said acquisition geometries specifying source and receiver locations, said method comprising: proposing two or more different acquisition geometries to evaluate; assuming a subsurface reference model of velocity or other physical property, simulating, using a computer, synthetic measured seismic data corresponding to all source and receiver locations in the two or more acquisition geometries; for each selected acquisition geometry, selecting from the synthetic measured seismic data those data corresponding to the source and receiver locations present in the selected acquisition geometry, and inverting the selected synthetic measured data by iterative, numerical inversion, using the computer, to obtain a final updated subsurface model; comparing the final updated model for each proposed acquisition geometry to the reference model using a selected quantitative measure of agreement with the reference model, then selecting an acquisition geometry by balancing a high quantitative measure of agreement with a low survey cost; carrying out a survey designed according to the selected acquisition geometry; wherein the two or more acquisition geometries are ranked in order of areal coverage, and the iterative numerical inversions are performed in order of smallest areal coverage to largest areal coverage, and the final updated model from the first inversion is used as a starting guess for the model for the second inversion, and wherein the iterative numerical inversion of the selected synthetic measured seismic data comprises selecting a starting model, then using the starting model to simulate predicted data, then quantitatively measuring degree of misfit between the predicted data and the synthetic measured data, then updating the starting model to reduce the degree of misfit, and repeating for a next iteration using the updated model.
2. The method of claim 1 , wherein the two or more selected acquisition geometries have a same receiver spacing and source-shot spacing, and a same range of source-receiver offsets.
3. The method of claim 1 , wherein the degree of misfit is quantitatively measured by a cross-correlation cost function.
4. The method of claim 1 , wherein the selected quantitative measure of agreement incorporates at least one of volume of investigation, accuracy of a final updated model, less misfit between the predicted data and the synthetic measured data, and rate of change in misfit improvement with iterations.
5. The method of claim 1 , wherein in the iterative, numerical inversion, a plurality of sources, or alternatively a plurality of receivers using source-receiver reciprocity, are encoded and inverted simultaneously with a single simulation to generate the predicted data.
6. The method of claim 5 , wherein the plurality of encoded sources are summed to form a composite gather, and the composite gather is simulated in the single simulation wherein predicted data are simulated for every receiver location in the composite gather.
7. The method of claim 6 , wherein at least one of the selected acquisition geometries involves some receiver locations that do not record some of the source excitations, and a cross-correlation cross function is used to measure the degree of misfit between the predicted data and the synthetic measured data.
8. The method of claim 1 , wherein the starting model is determined using a final model from a prior inversion with a different acquisition geometry or with a different reference model.
9. The method of claim 1 , wherein the selected quantitative measure incorporates volume of investigation, and the starting model is mi start and the updated model is mli nvert , then further comprising repeating the inversion with a second starting model m2 start to obtain a corresponding updated model m 2 i nvert , wherein the volume of investigation is computed as the ratio (miinve−m 2 in″vert)|(mistart−m 2 start).
10. The method of claim 1 , further comprising estimating a relative cost of performing a survey using each proposed acquisition geometry.
11. The method of claim 1 , wherein the two or more different acquisition geometries are different because of a difference in one or more of: source or receiver locations, source or receiver components, source or receiver density, source-to-receiver offsets and source-to-receiver azimuths.
12. The method of claim 1 , wherein the acquisition geometries are for a 3D survey, and the iterative, numerical inversion is full wavefield inversion.
13. The method of claim 1 , wherein the two or more acquisition geometries can be ranked in order of number of seismic traces that would be generated, from smallest to largest, and each acquisition geometry includes all traces present in all smaller acquisition geometries, plus at least one more, wherein a trace is defined by a particular source location and receiver location.
14. The method of claim 1 , further comprising prospecting for hydrocarbons using the results of the survey carried out.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
July 13, 2015
August 20, 2019
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.